elif dataset_name == 'test' or dataset_name == 'test_blur':
loadpath_dict = basedir + "..\\datasets\\preprocessed_HIG\\dataset_dict_" + dataset_name + ".pickle"
with open(loadpath_dict, 'rb') as f:
dataset_load = pickle.load(f)
##--bag of fingertip position
fingertip = {}
fingertip['hpe1_orig'] = []
fingertip['hig_hpe1'] = []
fingertip['hpe2'] = []
fingertip['hpe1_blur'] = []
fingertip['average'] = []
#--start
fusionnet.set_mode('eval')
ir_batch = np.zeros((1, 1, args.trainImageSize, args.trainImageSize),
dtype=np.float32)
depth_batch = np.zeros((1, 1, args.trainImageSize, args.trainImageSize),
dtype=np.float32)
vis = Visualize_combined_outputs(utils, 4, 1, camerawidth, cameraheight)
progressbar = trange(data_num, leave=True)
for i in progressbar:
#for i in range(data_num):
frame = i
##--input
if dataset_name == 'v1':
depth_orig = cv2.imread(
load_filepath_imgs + 'depth-%07d.png' % frame, 2)
f.write('traindataNum_VR20:%s\n' % traindataNum_vr20)
f.write('validateNum_uvr:%s\n' % validateNum)
f.close()
#--start
progress_train = progress.Progress(loss_names, pretrain=False)
progress_validate = progress.Progress(loss_names, pretrain=False)
iternum_train = traindataNum_uvr // args.train_batch
iternum_train_blur = traindataNum_blur_uvr // args.train_batch
iternum_vr20 = traindataNum_vr20 // args.train_batch
print('start..')
for epoch in range(args.epochs):
#--train
fusionnet.set_mode('train')
datasetloader_uvr['train'].shuffle()
generator_train_icvl = datasetloader_icvl.generator_learningData(
args.train_batch, 'train', False, 0)
#trange_num=traindataNum_uvr//args.train_batch + traindataNum_blur_uvr//args.train_batch +traindataNum_vr20//args.train_batch
if traindataNum_vr20 > 0:
trange_num = 2 * traindataNum_vr20 // args.train_batch
else:
trange_num = traindataNum_uvr // args.train_batch
progressbar = trange(trange_num, leave=True)
for i in progressbar:
#select dataset